As we currently await Irene, here are some personal observations regarding tropical storm forecast accuracy. 1) In my unsubstantiated experience, tropical storms seem to curve more to the east than the models predict. Irene has exhibited this behavior already. Several days ago it was forecast to come very close to Miami. Now it is a good 100 miles east of that track. 2) We should have some graphical representation of how inaccurate these models ( as summaraized and consolidated by the NHC official forecast track) were for a storm. You can argue the cone of probability serves this function now, but I would like to see both an actual storm path plot and say, a five day past forecast plot. It might be helpful to draw a vector from each 5 day past predicted plot point and date and time to the actual plot point for that 5 day past predicted date and time.
Sailors do pretty well at predicting paths based on which islands get buzzed along the way. I don't share your thoughts about curving to the East. Some years we get them all in the gulf and some years they go Atlantic. I really thought that the warms temps would have made this a nasty year for the gulf, but it just hasn't been so far. Still plenty of time for lots of damage to be visited in the gulf, but I would have guessed we have had at least several Cat 3's hit the SEC by now.
la nina has been in full effect this spring/summer. it will help in keeping most storms coming from the atlantic to the east coast this year.
So now I see the NHC does provide archived advisories and entire storm forecast animations according to arstechnica. The specific eastward curve drift I am refering to can be seen in the following NHC animation. Take specific note of the forecast path jsut east of Miami as it shifts continuously eastward as the actual path transits through and past the Bahamas. What I'd like to see added is something like a red line for the three day forecast position, a blue line for the two day forecast position, a green line for the one day forecast position and vectors from the plotted points to their corresponding actual points for the times predicted. Nonetheless the NHC archived advisory data should at least give me the data to create a few maps myself.
